822 



BELL SYSTEM TECHNICAL JOURNAL 



series of tests, in which several speakers were used, were then under- 

 taken on a carbon transmitter circuit. In Curve II the averages for 

 the two observers, of results on two voices, are shown. Three to 

 four weeks' time was spent by the observers in making the various 

 tests mentioned above. 



The curves under III show similar data that were taken at a later 

 date by one new observer, for several voices. All of the above tests 



100 



3 4 5 



NUMBER OF TEST 



Fig. 3 — Practice effects for an experienced crew 



were made with the standard lists. In Curve IV, data are shown 

 that were obtained with the vowel and consonant word lists (see 

 Table XVII). In these tests four new observers were used and no 

 preliminary training was given. It is evident that with the word 

 lists, the results reach a state of saturation very quickly. 



After a crew has spent several months in testing, its performance 

 becomes largely mechanical. Under such circumstances the practice 

 effects are rather small for types of distorted speech with which it 

 has had experience. When the speech distortion is unusual, however, 

 rather large practice effects may be obtained. Fig. 3 shows such 



